Dividend smoothing vs dividend signalling: evidence from UK firms

Purpose – We test for the validity of the smoothing and signalling hypotheses of dividend determination. Design/methodology/approach - Using a VAR framework we examine the dynamic behaviour of share prices, dividends and earnings for 137 UK manufacturing and service companies, observed over the period 1970-2003. Findings - There is strong evidence of a contemporaneous relationship between prices, dividends and earnings, and little evidence of independence between these variables. Some evidence in favour of both the smoothing and the signalling hypothesis is obtained from causality tests, with perhaps more support for the latter hypothesis. However, there is considerable diversity in the causal relationships between prices, dividends and earnings. Research limitations/implications - No single hypothesis regarding the determination of dividends, and the predictive power of dividends for earnings and prices appears to dominate. Originality/value - The results presented here are of interest to markets agents in that while they suggest there is no single transition mechanism linking prices, dividends and earnings, nevertheless these three variables are strongly correlated and exhibit varying degrees of causality.Antoregressive processes, Dividends, Vector